arXiv subjects

Fan Zhang

Publications and source records attributed to Fan Zhang.

At least 883 records · Page 49Linked to original sources

arXiv2017

Observation of topological valley transport of sound in sonic crystals

Valley pseudospin, labeling quantum states of energy extrema in momentum space, is attracting tremendous attention1-13 because of its potential in constructing new carrier of information. Compared with the non-topological bulk valley transport realized soon after predictions1-5, the topological valley transport in domain walls6-13 is extremely challenging owing to the inter-valley scattering inevitably induced by atomic scale imperfectness, until the recent electronic signature observed in bilayer graphene12,13. Here we report the first experimental observation of topological valley transport of sound in sonic crystals. The macroscopic nature of sonic crystals permits the flexible and accurate design of domain walls. In addition to a direct visualization of the valley-selective edge modes through spatial scanning of sound field, reflection immunity is observed in sharply curved interfaces. The topologically protected interface transport of sound, strikingly different from that in traditional sound waveguides14,15, may serve as the basis of designing devices with unconventional functions.

cond-mat.mes-hall↗

arXiv2017

Circular Dichroism and Radial Hall Effects in Topological Materials

Under symmetry breaking, a three-dimensional nodal-line semimetal can turn into a topological insulator or Weyl semimetal, accompanied by the generation of momentum-space Berry curvature. We develop a theory that unifies their circular dichroism and highlights the roles of Berry curvature distribution and light incident direction. Nontrivially, these phases exhibit distinct dichroic optical absorption and radial Hall effects, with characteristic scalings with photon energy and electric field. Our findings offer a new diagnosis tool for examining topological phases of matter.

cond-mat.mtrl-sci↗

arXiv2017

Doubly Robust Data-Driven Distributionally Robust Optimization

Data-driven Distributionally Robust Optimization (DD-DRO) via optimal transport has been shown to encompass a wide range of popular machine learning algorithms. The distributional uncertainty size is often shown to correspond to the regularization parameter. The type of regularization (e.g. the norm used to regularize) corresponds to the shape of the distributional uncertainty. We propose a data-driven robust optimization methodology to inform the transportation cost underlying the definition of the distributional uncertainty. We show empirically that this additional layer of robustification, which produces a method we called doubly robust data-driven distributionally robust optimization (DD-R-DRO), allows to enhance the generalization properties of regularized estimators while reducing testing error relative to state-of-the-art classifiers in a wide range of data sets.

stat.ML↗

arXiv2017

Application of Finite Element, Phase-field, and CALPHAD-based Methods to Additive Manufacturing of Ni-based Superalloys

Numerical simulations are used in this work to investigate aspects of microstructure and microsegregation during rapid solidification of a Ni-based superalloy in a laser powder bed fusion additive manufacturing process. Thermal modeling by finite element analysis simulates the laser melt pool, with surface temperatures in agreement with in situ thermographic measurements on Inconel 625. Geometric and thermal features of the simulated melt pools are extracted and used in subsequent mesoscale simulations. Solidification in the melt pool is simulated on two length scales. For the multicomponent alloy Inconel 625, microsegregation between dendrite arms is calculated using the Scheil-Gulliver solidification model and DICTRA software. Phase-field simulations, using Ni-Nb as a binary analogue to Inconel 625, produced microstructures with primary cellular/dendritic arm spacings in agreement with measured spacings in experimentally observed microstructures and a lesser extent of microsegregation than predicted by DICTRA simulations. The composition profiles are used to compare thermodynamic driving forces for nucleation against experimentally observed precipitates identified by electron and X-ray diffraction analyses. Our analysis lists the precipitates that may form from FCC phase of enriched interdendritic compositions and compares these against experimentally observed phases from 1 h heat treatments at two temperatures: stress relief at 1143 K (870°C) or homogenization at 1423 K (1150°C).

cond-mat.mtrl-sci↗

arXiv2017

Topological, Valleytronic, and Optical Properties of Monolayer PbS

We systematically examine the topological, valleytronic, and optical properties of experimentally accessible PbS (001) few-layers, with a focus on the monolayer. With even-odd layer-dependent oscillations and without band inversions, the band gaps cover a wide spectrum from infrared to visible, making the few-layers promising for optoelectronics. Intriguingly, the uniaxial (biaxial) compressive strain can tune the trivial monolayer into a $\mathcal{Z}_2$-topological (topological crystalline) insulator, which may be utilized for controllable low-power electronic devices. Although elliptical dichroism vanishes in the monolayer due to inversion symmetry, optical pumping provides an efficient tool to characterize the three phases and to realize charge, spin, and valley Hall effects in optoelectronic transport that are tunable by external strain and light ellipticity.

cond-mat.mes-hall↗

arXiv2017

A Joint Identification Approach for Argumentative Writing Revisions

Prior work on revision identification typically uses a pipeline method: revision extraction is first conducted to identify the locations of revisions and revision classification is then conducted on the identified revisions. Such a setting propagates the errors of the revision extraction step to the revision classification step. This paper proposes an approach that identifies the revision location and the revision type jointly to solve the issue of error propagation. It utilizes a sequence representation of revisions and conducts sequence labeling for revision identification. A mutation-based approach is utilized to update identification sequences. Results demonstrate that our proposed approach yields better performance on both revision location extraction and revision type classification compared to a pipeline baseline.

cs.CL↗

arXiv2017

A multi-task convolutional neural network for mega-city analysis using very high resolution satellite imagery and geospatial data

Mega-city analysis with very high resolution (VHR) satellite images has been drawing increasing interest in the fields of city planning and social investigation. It is known that accurate land-use, urban density, and population distribution information is the key to mega-city monitoring and environmental studies. Therefore, how to generate land-use, urban density, and population distribution maps at a fine scale using VHR satellite images has become a hot topic. Previous studies have focused solely on individual tasks with elaborate hand-crafted features and have ignored the relationship between different tasks. In this study, we aim to propose a universal framework which can: 1) automatically learn the internal feature representation from the raw image data; and 2) simultaneously produce fine-scale land-use, urban density, and population distribution maps. For the first target, a deep convolutional neural network (CNN) is applied to learn the hierarchical feature representation from the raw image data. For the second target, a novel CNN-based universal framework is proposed to process the VHR satellite images and generate the land-use, urban density, and population distribution maps. To the best of our knowledge, this is the first CNN-based mega-city analysis method which can process a VHR remote sensing image with such a large data volume. A VHR satellite image (1.2 m spatial resolution) of the center of Wuhan covering an area of 2606 km2 was used to evaluate the proposed method. The experimental results confirm that the proposed method can achieve a promising accuracy for land-use, urban density, and population distribution maps.

cs.CV↗

arXiv2017

A vertex-weighted-Least-Squares gradient reconstruction

Gradient reconstruction is a key process for the spatial accuracy and robustness of finite volume method, especially in industrial aerodynamic applications in which grid quality affects reconstruction methods significantly. A novel gradient reconstruction method for cell-centered finite volume scheme is introduced. This method is composed of two successive steps. First, a vertex-based weighted-least-squares procedure is implemented to calculate vertex gradients, and then the cell-centered gradients are calculated by an arithmetic averaging procedure. By using these two procedures, extended stencils are implemented in the calculations, and the accuracy of gradient reconstruction is improved by the weighting procedure. In the given test cases, the proposed method is showing improvement on both the accuracy and convergence. Furthermore, the method could be extended to the calculation of viscous fluxes.

physics.flu-dyn↗

arXiv2017

Pulsar magnetospheric convulsions induced by an external magnetic field

The canonical pulsar magnetosphere contains a bubble of closed magnetic field lines that is separated from the open lines by current sheets, and different branches of such sheets intersect at a critical line on the light cylinder (LC). The LC is located far away from the neutron star, and the pulsar's intrinsic magnetic field at that location is much weaker than the commonly quoted numbers applicable to the star surface. The magnetic field surrounding supermassive black holes that reside in galactic nuclei is of comparable or greater strength. Therefore, when the pulsar travels inside such regions, a non-negligible Lorentz force is experienced by the current sheets, which tends to pull them apart at the critical line. As breakage occurs, instabilities ensue that burst the bubble, allowing closed field lines to snap open and release large amounts of electromagnetic energy, sufficient to power fast radio bursts (FRBs). This process is necessarily associated with an environment of a strong magnetic field and thus might explain the large rotation measures recorded for the FRBs. We sketch a portrait of the process and examine its compatibility with several other salient features of the FRBs.

astro-ph.HE↗

arXiv2017

Odd-integer quantum Hall states and giant spin susceptibility in p-type few-layer WSe2

We fabricate high-mobility p-type few-layer WSe2 field-effect transistors and surprisingly observe a series of quantum Hall (QH) states following an unconventional sequence predominated by odd-integer states under a moderate strength magnetic field. By tilting the magnetic field, we discover Landau level (LL) crossing effects at ultra-low coincident angles, revealing that the Zeeman energy is about three times as large as the cyclotron energy near the valence band top at Γ valley. This result implies the significant roles played by the exchange interactions in p-type few-layer WSe2, in which itinerant or QH ferromagnetism likely occurs. Evidently, the Γ valley of few-layer WSe2 offers a unique platform with unusually heavy hole-carriers and a substantially enhanced g-factor for exploring strongly correlated phenomena.

cond-mat.mes-hall↗

arXiv2016

Observation of Acoustic Valley Vortex States and Valley-Chirality Locked Beam Splitting

The Letter reports an experimental observation of the classical version of valley polarized states in a two-dimensional hexagonal sonic crystal, where the inversion-symmetry breaking of scatterers induces an omnidirectional frequency gap. The acoustic valley states, which carry specific linear momenta and orbital angular momenta, were selectively excited by external Gaussian beams and conveniently confirmed by the pressure distribution outside the crystal, according to the criterion of momentum conservation. The vortex nature of such intriguing crystal states was directly characterized by scanning the phase profile inside the crystal. In addition, we observed a peculiar beam splitting phenomenon, in which the separated beams are constructed by different valleys and locked to the opposite vortex chirality. The exceptional sound transport, encoded with valley-chirality locked information, may serve as the basis of designing conceptually novel acoustic devices with unconventional functions.

cond-mat.mes-hall↗

arXiv2016

Towards an understanding of the force-free magnetosphere of rapidly spinning black holes

The ability of a plasma surrounding spinning black holes to extract rotational energy and power energetic emissions has been recognized as a key astrophysical phenomenon. Important insights into the nature of this process are obtained through the analysis of the interplay between a force-free magnetosphere and the black hole. This task involves solving a complicated system of equations, often requiring complex numerical simulations. Recent analytical attempts at tackling this problem have exploited the fact that the near horizon region of extreme Kerr (NHEK) is endowed with an enhanced symmetry group. We continue in this direction and show that for some conformally self-similar solutions, the NHEK force-free equations reduce to a single non-linear ordinary differential equation which is difficult to solve with straightforward integration. We here introduce a new approach specifically tailored to this problem and describe how one can obtain physically meaningful solutions.

astro-ph.HE↗

arXiv2016

Magnetic brightening and control of dark excitons in monolayer WSe2

Monolayer transition metal dichalcogenide (TMDC) crystals, as direct-gap materials with unusually strong light-matter interaction, have attracted much recent attention. In contrast to the initial understanding, the minima of the conduction band are predicted to be spin split. Because of this splitting and the spin-polarized character of the valence bands, the lowest-lying excitonic states in WX2 (X=S, Se) are expected to be spin-forbidden and optically dark. To date, however, there has been no direct experimental probe of these dark band-edge excitons, which strongly influence the light emission properties of the material. Here we show how an in-plane magnetic field can brighten the dark excitonic states and allow their properties to be revealed experimentally in monolayer WSe2. In particular, precise energy levels for both the neutral and charged dark excitons were obtained and compared with ab-initio calculations using the GW-BSE approach. Greatly increased emission and valley lifetimes were observed for the brightened dark states as a result of their spin configuration. These studies directly probe the excitonic spin manifold and provide a new route to tune the optical and valley properties of these prototypical two-dimensional semiconductors.

cond-mat.mes-hall↗

arXiv2016

Representation Learning with Deconvolution for Multivariate Time Series Classification and Visualization

We propose a new model based on the deconvolutional networks and SAX discretization to learn the representation for multivariate time series. Deconvolutional networks fully exploit the advantage the powerful expressiveness of deep neural networks in the manner of unsupervised learning. We design a network structure specifically to capture the cross-channel correlation with deconvolution, forcing the pooling operation to perform the dimension reduction along each position in the individual channel. Discretization based on Symbolic Aggregate Approximation is applied on the feature vectors to further extract the bag of features. We show how this representation and bag of features helps on classification. A full comparison with the sequence distance based approach is provided to demonstrate the effectiveness of our approach on the standard datasets. We further build the Markov matrix from the discretized representation from the deconvolution to visualize the time series as complex networks, which show more class-specific statistical properties and clear structures with respect to different labels.

cs.LG↗

arXiv2016

When Engagement Meets Similarity: Efficient (k,r)-Core Computation on Social Networks

In this paper, we investigate the problem of (k,r)-core which intends to find cohesive subgraphs on social networks considering both user engagement and similarity perspectives. In particular, we adopt the popular concept of k-core to guarantee the engagement of the users (vertices) in a group (subgraph) where each vertex in a (k,r)-core connects to at least k other vertices. Meanwhile, we also consider the pairwise similarity between users based on their profiles. For a given similarity metric and a similarity threshold r, the similarity between any two vertices in a (k,r)-core is ensured not less than r. Efficient algorithms are proposed to enumerate all maximal (k,r)-cores and find the maximum (k,r)-core, where both problems are shown to be NP-hard. Effective pruning techniques significantly reduce the search space of two algorithms and a novel (k,k')-core based (k,r)-core size upper bound enhances performance of the maximum (k,r)-core computation. We also devise effective search orders to accommodate the different nature of two mining algorithms. Comprehensive experiments on real-life data demonstrate that the maximal/maximum (k,r)-cores enable us to find interesting cohesive subgraphs, and performance of two mining algorithms is significantly improved by proposed techniques.

cs.SI↗

arXiv2016

A global optimization algorithm for sparse mixed membership matrix factorization

Mixed membership factorization is a popular approach for analyzing data sets that have within-sample heterogeneity. In recent years, several algorithms have been developed for mixed membership matrix factorization, but they only guarantee estimates from a local optimum. Here, we derive a global optimization (GOP) algorithm that provides a guaranteed $ε$-global optimum for a sparse mixed membership matrix factorization problem. We test the algorithm on simulated data and find the algorithm always bounds the global optimum across random initializations and explores multiple modes efficiently.

stat.ME↗

arXiv2016

Long distance co-propagation of quantum key distribution and terabit classical optical data channels

Quantum key distribution (QKD) generates symmetric keys between two remote parties, and guarantees the keys not accessible to any third party. Wavelength division multiplexing (WDM) between QKD and classical optical communications by sharing the existing fibre optics infrastructure is highly desired in order to reduce the cost of QKD applications. However, quantum signals are extremely weak and thus easily affected by the spontaneous Raman scattering effect from intensive classical light. Here, by means of wavelength selecting and spectral and temporal filtering, we realize the multiplexing and long distance co-propagation of QKD and Terabit classical coherent optical communication system up to 80km. The data capacity is two orders of magnitude larger than the previous results. Our demonstration verifies the feasibility of QKD and classical communication to share the resources of backbone fibre links, and thus taking the utility of QKD a great step forward.

quant-ph↗

arXiv2016

Indoor Space Recognition using Deep Convolutional Neural Network: A Case Study at MIT Campus

In this paper, we propose a robust and parsimonious approach using Deep Convolutional Neural Network (DCNN) to recognize and interpret interior space. DCNN has achieved incredible success in object and scene recognition. In this study we design and train a DCNN to classify a pre-zoning indoor space, and from a single phone photo to recognize the learned space features, with no need of additional assistive technology. We collect more than 600,000 images inside MIT campus buildings to train our DCNN model, and achieved 97.9% accuracy in validation dataset and 81.7% accuracy in test dataset based on spatial-scale fixed model. Furthermore, the recognition accuracy and spatial resolution can be potentially improved through multiscale classification model. We identify the discriminative image regions through Class Activating Mapping (CAM) technique, to observe the model's behavior in how to recognize space and interpret it in an abstract way. By evaluating the results with misclassification matrix, we investigate the visual spatial feature of interior space by looking into its visual similarity and visual distinctiveness, giving insights into interior design and human indoor perception and wayfinding research. The contribution of this paper is threefold. First, we propose a robust and parsimonious approach for indoor navigation using DCNN. Second, we demonstrate that DCNN also has a potential capability in space feature learning and recognition, even under severe appearance changes. Third, we introduce a DCNN based approach to look into the visual similarity and visual distinctiveness of interior space.

cs.CV↗